Gain-of-function mutations in the viral dsRNA sensor melanoma differentiation–associated protein 5 (MDA5) lead to autoimmune IFNopathies, including Singleton–Merten syndrome (SMS) and Aicardi–Goutières syndrome. However, much remains unclear regarding the mechanism of disease progression and how external factors such as infection or immune stimulation with vaccination can affect the immune response. With this aim, we generated mice with human MDA5 bearing the SMS-associated mutation R822Q (hM-R822Q). hM-R822Q transgenic (Tg) mice developed SMS-like heart fibrosis, aortic valve enlargement, and aortic calcification with a systemic IFN-stimulated gene signature resulting in the activation of the adaptive immune response. Although administration of the viral dsRNA mimic polyinosinic-polycytidylic acid [poly(I:C)] did not have remarkable effects on the cardiac phenotype, dramatic inflammation was observed in the intestines where IFN production was most elevated. Poly(I:C)-injected hM-R822Q Tg mice also developed lethal hypercytokinemia marked by massive IL-6 levels in the serum. Interrupting the IFN signaling through mitochondrial antiviral signaling protein or IFN-α/β receptor alleviated hM-R822Q–induced inflammation. Furthermore, inhibition of JAK signaling with tofacitinib reduced cytokine production and ameliorated mucosal damage, enabling the survival of poly(I:C)-injected hM-R822Q Tg mice. These findings demonstrate that the MDA5 R822Q mutant introduces a critical risk factor for uncontrollable inflammation on viral infection or vaccination.
Singleton–Merten syndrome (SMS) is an autosomal dominant disease that presents primarily as skeletal, dental, and vascular abnormalities (1, 2). Calcifications of the aorta, aortic valve, and mitral valve have been widely observed in SMS cases (3, 4). Less common features include generalized muscle weakness, psoriasis, early-onset glaucoma, as well as recurrent infections (5, 6). SMS has been associated with mutations in the IFN induced with helicase C domain 1 gene (IFIH1), which encodes melanoma differentiation–associated protein 5 (MDA5), and the DExD/H-box helicase 58 gene (DDX58), which encodes retinoic acid–inducible gene I (RIG-I). MDA5 and RIG-I, together with Laboratory of Genetics and Physiology 2, belong to a family of cytosolic dsRNA receptors known as RIG-I–like receptors (RLRs). Their ability to recognize RNA viruses, as well as RNA intermediates of DNA virus transcription, make RLRs key effectors in the antiviral response. Binding to dsRNA activates MDA5 and RIG-I, which then interact with the adaptor protein mitochondrial antiviral signaling (MAVS). MAVS then signals through IFN regulatory factor 3 and NF-κB, leading to the production of proinflammatory responses, including the expression of IFN-β and proinflammatory cytokines such as IL-6 and IL-1β. Further downstream, IFN-β binds the IFN-α/β receptor (IFNAR), inducing the transcriptional upregulation of IFN-stimulated genes (ISGs) (7).
Whole-exome sequencing of DNA from families of SMS patients allowed the identification of a c.2465G>A (p.R822Q) missense mutation in IFIH1 within the region that encodes the HEL2 helicase domain of MDA5, close to the ATP binding site. This mutation potentially alters the stability of MDA5 binding to nucleic acids, leading to sustained MDA5 activation. Overexpressing MDA5 R822Q in HEK293T cells resulted in MDA5 hyperactivity when challenged with a dsRNA analog, poly(I:C) (3). RIG-I mutations c.1118A>C (p.E373A) and c.803G>T (p.C268F) have also been shown to cause atypical SMS wherein patients exhibited variable symptoms without dental abnormalities (8).
Although the mechanism of disease development in SMS patients remains unclear, type I IFN–driven chronic inflammation has been found to be the central factor in the development of SMS-associated symptoms. IL-6 is a major factor in chronic inflammatory diseases, autoimmune diseases, cytokine release syndrome, and cancer (9). IL-6 overexpression has also been associated with MDA5-mediated psoriasis (10). In addition, IL-6 is involved in the uncontrollable and rapid progression of inflammation typically resulting from infection (11, 12). In a physiological reaction referred to as the cytokine storm, the massive production of cytokines, chemokines, and growth factors contributes to secondary organ damage (13). Serum levels of TNF-α and IL-6 are found to be reliable predictors of disease severity and death (14, 15), whereas JAK inhibition and IL-6 blockade can reduce disease severity by lowering the levels of these cytokines (16, 17).
In mice, N-ethyl-N-nitrosourea–induced p.Glu821Ser substitution in Ifih1 leads to chronic activation of MDA5. Mice harboring the single missense mutation G821S in Ifih1 (Ifih1gs/+) develop lupus-like, SMS-like, and some Aicardi–Goutières syndrome–linked symptoms, including nephritis, dermatitis, abnormal bone development, and encephalitis (18–20). These mice had upregulated expression of IFN-inducible genes and inflammatory cytokines. In contrast, mice with human RIG-I E373A developed psoriatic skin lesions (21). In these mice, long-term inhibition of JAK1 and JAK3 signaling promoted healing, and pretreatment from an early age prevented lesion development. In addition, IL-17a–deficient hRIG-I E373A transgenic (Tg) mice did not develop skin lesions, linking Th17 and the psoriasis phenotype (21). These observations underscore the notion that RLR signaling impacts numerous cell types. The effects of RLR dysregulation are thus expansive, leading to a plethora of disease symptoms. Consequently, effective treatment is heavily reliant on better understanding of RLR dysregulation and the underlying mechanism.
Despite the availability of genetic data, differences in clinical symptoms in patients with similar gain-of-function mutations expose gaps in our knowledge of the development of associated IFNopathies. Previous studies on Ifih1gs/+ mice resulted in severe systemic disorder, with mortality typically observed in the first 8 wk even in this heterogenous genotype (18, 19). This significantly limited phenotypic and drug treatments that could be performed or administered. To closely examine the specific effects of the SMS-associated MDA5 mutation in vivo, we introduced a human MDA5 R822Q mutant transgene into mice through bacterial artificial chromosome (BAC) insertion. Notably, research by Crampton et al. (22) has shown that the BAC insertion of 12 copies of wild-type (WT) murine Ifih1 resulted in levels of circulating cytokines and no organ pathological abnormalities despite having elevated MDA5, IFN, and ISG expression.
The inflammation-induced disease phenotype was monitored in these mice. In addition, poly(I:C), which is known to induce signaling through MDA5 and TLR3 (23, 24), was introduced to mimic active inflammation triggered by infection. We demonstrated that nonpathogenic autoimmune activity at the basal level predisposed hMDA5 R822Q Tg mice to a severe inflammatory response on immune stimulation. We confirmed the roles of MAVS and IFN receptor signaling in MDA5 R822Q-mediated disease progression and introduced a potential therapeutic treatment to prevent pathological immune hyperactivity.
Materials and Methods
Mice and treatments
BAC technology was used to clone a fragment of the human chromosome encompassing the IFIH1 gene together with its inherent promoter. This IFIH1 gene was mutated to encode MDA5 R822Q and transferred by random insertion into C57BL/6J mice (Institute of Immunology Co., Tokyo, Japan). PCR-based genotyping of human MDA5 bearing the SMS-associated mutation R822Q (hM-R822Q) Tg mice was performed using the forward primer 5′-TTCAGGCTTCCTAAGCTCG-3′ and reverse primer 5′-GAGTCAATGACACAAATGCC-3′, which targeted exon 5 of human IFIH1, including the site of the c.2465G>A missense mutation. All hM-R822Q Tg mice used in these experiments were heterozygous for the human MDA5 mutation and originated from one line that was confirmed to show germline transmission. Mouse lines were crossed to generate Ifnar1−/− and Mavs−/− hM-R822Q mice. Ifnar1−/− mice were purchased from B&K Universal. Mavs−/− mice were a kind gift from S. Akira (Osaka University). Mice were analyzed at 8–10 wk of age unless otherwise specified. Ifih1gs/− mice were bred with C57BL/6J background WT mice by in vitro fertilization. All mice were housed in a specific pathogen-free facility.
Poly(I:C) (GE Healthcare) was injected i.p. at 15 μg/g diluted to 500 μl with PBS per mouse. Tofacitinib solution was prepared by dissolving CP 690550 citrate (TOCRIS Bioscience) in a solution of 0.5% methylcellulose (Wako)/0.025% Tween 20 (Nacalai Tesque). Mice were administered with 0.3 mg/10 g tofacitinib (200 μl) per mouse through oral gavage. All poly(I:C) injection and tofacitinib treatment experiments were performed on female mice unless otherwise specified.
Tissue samples were fixed in 4% paraformaldehyde phosphate (Nacalai Tesque) and embedded in paraffin. Sections (5 μm) were stained with H&E (Sakura). Picrosirius red staining was performed on heart samples using Direct Red 80 (Sigma-Aldrich) in saturated picric acid solution and counterstained with hematoxylin. Stained sections were observed under a BZ-8000 microscope (KEYENCE). Immunofluorescence staining was performed on 2-μm sections, and images were captured with a confocal microscope (TCS SP8; Leica). Histological analysis of tissues was performed through blind evaluation. Numerical scores for the intestines were based on villous lifting, villous shortening, epithelial shedding, and cellular infiltration (localized or extensive).
RNA isolation and quantification
Total RNA was extracted from homogenized organ samples using the TRIzol reagent (Invitrogen, Life Technologies) and transcribed using ReverTra AceR qPCR RT Master Mix with gDNA Remover (TOYOBO). Real-time PCR was performed with Thunderbird SYBR qPCR mix (TOYOBO) on a StepOnePlus Real-Time PCR system (Applied Biosystems). Target expression was normalized to an endogenous reference, GAPDH.
Aortic calcium was measured with a Colorimetric Calcium Detection Assay Kit (Abcam) following the manufacturer’s instructions. In brief, aortas were collected and washed in cold PBS. Tissues were suspended in calcium assay buffer and homogenized using a sonicator. After centrifugation, 50 μl of the sample supernatant was mixed with 90 μl of chromogenic agent and 60 μl calcium assay buffer. After 5–10 min, the absorbance at 575 nm of each sample solution was read.
Plasma IL-6 levels were measured using an ELISA MAX Deluxe Set Mouse IL-6 kit (BioLegend). Uncoated flat-bottom 96-well plates were coated with IL-6 Ab overnight at 4°C. Wells were washed before blocking for 1 h at room temperature. After washing, samples were added to the wells and incubated at room temperature for 2 h. The wells were washed again, and the Detection Ab solution (100 μl) was added to the wells and incubated for another hour at room temperature. After washing, 100 μl Avidin-HRP solution was added to each well. After 30 min at room temperature, the wells were washed carefully. The tetramethylbenzidine substrate solution (100 μl) was added to the wells and kept in the dark for 20 min before the addition of stop solution. Absorbance was read at 450 nm with wavelength correction at 570 nm.
Cell isolation and flow cytometry
Spleens were harvested and cellularized by mechanical dissociation through a 70-μm filter. Splenocytes were treated with ammonium-chloride-potassium lysis buffer. Surface Ags were stained with fluorochrome-conjugated Abs. Dendritic cells were stained after CD19- and CD3e-based depletion using MACS cell separation (Miltenyi Biotec). Intracellular staining was performed using a Foxp3 Fixation/Permeabilization kit (eBioscience). For IL-6 and IFN-γ staining, cells were first stimulated with PMA/ionomycin (Sigma-Aldrich) or 1 µg/ml LPS (Escherichia coli O111:B4; Sigma-Aldrich, St. Louis, MO) for 4–6 h at 37°C. Monensin (1:1000) was added after 1 h of incubation. Cells were stained with Abs conjugated with Alexa Fluor 488 for CD44 (103016; BioLegend), CD69 (104516; BioLegend), and CD86 (105108; BioLegend); PE for CD3e (100308; BioLegend), CD11b (101208; BioLegend), CD11c (117308; BioLegend), CD62L (561918; Pharmingen, BD), and IFN-γ (504507; BioLegend); allophycocyanin for CD8a (100732; BioLegend), CD19 (115512; BioLegend), CD45R/B220 (553092; Pharmingen, BD), IL-6 (504507; BioLegend), and NK1.1 (108709; BioLegend); and PE/Cyanine7 (100528; BioLegend). Stained cells were detected using a BD LSRFortessa X20 (BD Biosciences) system, and the data were analyzed with the FlowJo software (Tomy Digital Biology).
Bone phenotype analysis
Femurs of 15- to 16-wk-old male mice were fixed in 70% ethanol. These samples were subjected to microcomputed tomography and scanned using a ScanXmate-L090H (Comscantecno) system as described by Kurotaki et al. (25). Reconstruction and analysis of the three-dimensional microstructural image data were performed using coneCTexpress (WhiteRabbit) and TRI/3DBON-FCS (RATOC Systems) systems.
In vitro differentiation of osteoclasts
Bone marrow cells were obtained by flushing the femurs and tibiae of mice. Cells were suspended in RPMI 1640 (Nacalai Tesque) containing penicillin-streptomycin (100 U/ml and 100 mg/ml, respectively; Nacalai Tesque) and 10% FBS (Life Technologies) supplemented with 10 ng/ml M-CSF (R&D Systems). After 3 d, cultures were rinsed with PBS, and the attached bone marrow–derived macrophages were collected by treatment with 0.5% trypsin-EDTA (Nacalai Tesque). Bone marrow–derived macrophages were cultured in medium containing 10 ng/ml M-CSF (R&D Systems) and 40 ng/ml receptor activator for NF-κB ligand (R&D Systems) for 4 d, with medium change on the second day. The differentiated osteoclasts were fixed with 4% paraformaldehyde and stained using a telomerase repeat amplification protocol (TRAP). Quantification of osteoclast differentiation was carried out indirectly by measuring TRAP activity in cell lysates using a TRAP and alkaline phosphatase assay (Takara Bio).
Statistical analyses were performed using GraphPad Prism, version 9.0 (GraphPad Software, San Diego, CA). Differences between two experimental groups were analyzed using a two-tailed t test. Multiple comparisons were performed using a one-way ANOVA test. A p value <0.05 was considered statistically significant (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ***p ≤ 0.0001).
All animal experiments were performed following regulations approved by the Committee for Animal Experiments of the Institute for Frontier Life and Medical Sciences of Kyoto University.
SMS-like pathogenesis in hM-R822Q MDA5 Tg mice
To gain a better understanding of the role of R822Q MDA5 in SMS in vivo, we generated Tg mice through the insertion of a BAC carrying the mutated genomic human MDA5 into the C57BL/6J genome. Mice showed signs of low fertility with infrequent pregnancies and small litters. Female hM-R822Q Tg mice also often died midlabor. Although no clear evidence of fatal disease was observed, both male and female hM-R822Q mice lived an average of 80 wk (Fig. 1A), whereas WT mice lived for >140 wk in the same housing conditions. This indicates a shortened life span in hM-R822Q Tg mice. Body weights taken at 10 wk showed a trend for decreased body weight of hM-R822Q Tg mice compared with WT littermates, demonstrating early signs of growth retardation (Fig. 1B). In addition, hM-R822Q Tg mice developed abnormalities in spine curvature noticeable from as early as 14 wk (Supplemental Fig. 1C). Despite the absence of observable irregularities in femoral bone density in comparison with WT mice (Supplemental Fig. 1D) and lack of significant difference in bone mineral density (Supplemental Fig. 1E), there were indications of impaired osteoclast differentiation. Bone marrow macrophages from hM-R822Q Tg mice induced for osteoclastogenesis with M-CSF and receptor activator for NF-κB ligand showed decreased formation of multinucleated TRAP-positive cells compared with WT cells (Supplemental Fig. 1F).
In childhood, SMS patients have been found to develop calcifications in the aorta, as well as the aortic and mitral valves, which, in some cases, lead to early death (3). Inflammation has been known to put a strain on the heart, heart valves, and blood vessels, inducing damage (26). hM-R822Q Tg mice displayed subtle signs of valve degeneration and valve enlargement and increased levels of cardiac fibrosis (Fig. 1C). Calcium content of the aortas of hM-R822Q Tg mice were significantly elevated compared with WT mice (Fig. 1D). The RNA expression profile of the heart showed activated type I IFN signaling, including elevated levels of Ifnb and Isg56 (Fig. 1E). Among the cytokines found to be upregulated in hM-R822Q Tg hearts were Il6 and Tnfa, two cytokines known to be increased in calcified aortic valve disease (27). No difference was observed for regulatory cytokines such as Tgfb and Il10 (Fig. 1F). A similar pattern could be observed in the aorta in which RNA levels of Ifnb, Il6, Isg56, and Tnfa showed increased expression (Fig. 1G).
In summary, hM-R822Q mice displayed an inflammatory cytokine expression profile that is conducive to calcification in the aortic valves and the aorta. Increased levels of aortic calcium, heart fibrosis, and mild valve degeneration showed a strong similarity to SMS-like cardiac calcifications. Together with impaired osteoclast development and early progression of bone abnormalities, the hM-R822Q Tg mice displayed SMS-associated symptoms.
hM-R822Q Tg mice display systemic inflammation
To understand the systemic effects of constitutively active MDA5 in mice, we conducted a multiorgan assessment of inflammation and inflammation-induced damage. All investigated organs in hM-R822Q Tg mice showed upregulation of inflammatory cytokines and ISGs such as Ifnb and Isg56 (Fig. 2A). IFN levels were particularly elevated in mucosal regions, such as the intestines, lungs, and eyes. The spleens of hM-R822Q mice were enlarged compared with WT mice (Fig. 2B). Splenic macrophages (F4/80+CD11b+), dendritic cells (CD11c+), and NK cells (NK1.1+CD3e−) showed increased expression of cell-specific CD86 and CD69 activation markers in hM-R822Q Tg mice compared with WT mice, despite having no marked differences in the percentages of their splenic immune cell populations (Fig. 2C). The hM-R822Q Tg mice also displayed a primed lymphocyte response in the absence of active infection. The T cells (CD19+) and B cells (CD3e+) in Tg mice had higher CD69 expression (Fig. 2D) than that of WT mice, with increased populations of effector CD4+ and CD8+ T cells (Fig. 2E). This indicates that adaptive immunity was also significantly activated.
Overall, hM-R822Q Tg mice exhibited systemic inflammation with mild pathology, resulting in a short life span. Furthermore, cytokine expression and cell activity indicated a high risk for the occurrence of inflammation-associated organ damage.
Induced lethal inflammation in female hM-R822Q Tg mice
We next assessed what phenotype may be induced in the mutant mice using poly(I:C), which mimics virus infection. Male hM-R822Q mice i.p. injected with poly(I:C) did not show indications of significant disease phenotype (Supplemental Fig. 2). In contrast, female mice injected with poly(I:C) displayed inflammation in the eye region with mucosal discharge (Supplemental Fig. 3A, 3B) and diarrhea within 16 h postinjection. Poly(I:C)-treated female hM-R822Q mice exhibited severe weakness, and 75% of these mice died within 22 h postinjection (Fig. 3A). Both WT and hM-R822Q female mice lost weight after poly(I:C) injection, with hM-R822Q female mice displaying a significantly higher percentage of weight loss relative to their weights at baseline (Fig. 3B).
In the poly(I:C)-injected mice, symptoms of diarrhea were indicative of intestinal inflammation. The poly(I:C)-injected hM-R822Q mice showed visible intestinal distension not observed in WT mice (Supplemental Fig. 3B). Severe villous shortening was observed in the small intestine of the poly(I:C)-injected Tg mice, and both small and large intestines of the Tg mice displayed cellular infiltration not observed in poly(I:C)-injected WT mice (Fig. 3C).
The symptoms and gene expression profiles of poly(I:C)-injected hM-R822Q Tg mice suggested a massive inflammatory response. We checked markers of cytokine overproduction in the ileum and colon and found that Il6 mRNA was upregulated in the hM-R822Q mice compared with that in the WT mice (Fig. 3D, 3E). mRNA levels of Il1b, a cytokine that is associated with increased permeability of the intestinal barrier in inflammatory bowel disease (IBD) (28, 29), was increased in poly(I:C)-injected hM-R822Q mice. Tnfa, which is known to be increased in IBD (30, 31), was also elevated in the intestines of hM-R822Q Tg mice. Larger populations of IL-6–producing CD4+ cells were found in the spleen of hM-R822Q mice after poly(I:C) injection (Fig. 3F). Poly(I:C) injection concurrently induced a drastic spike in serum IL-6 levels in hM-R822Q Tg mice compared with nontreated WT mice (Fig. 3G).
The liver also showed signs of cell infiltration after poly(I:C) injection (Supplemental Fig. 3C). Analysis of cytokine mRNA expression in the liver showed drastic upregulation, particularly of Ifnb, Il6, and Il1b (Supplemental Fig. 3D), raising the possibility of secondary organ damage, which is often associated with lethal inflammation because of IFN and cytokine overproduction.
Taken together, we found that injection with the MDA5 activator poly(I:C) caused massive inflammation in female hM-R822Q Tg mice that can cause death within the first 24 h. We believe this to be the result of a combination of hypercytokinemia, cell infiltration in the organs, and inflammation-induced damage to the mucosa.
MAVS- and IFNAR-dependent inflammation and poly(I:C)-induced pathology
Random insertion of the mutant hMDA5 transgene in mice called into question whether the observed phenotype was a direct result of hM-R822Q signaling, aberrant signaling by the transgene, or even a by-product of the insertion. In addition, poly(I:C) is also able to trigger other dsRNA sensors, such as TLR3. MAVS signals downstream of RLRs, but not TLR3. In contrast, IFNAR binds to type I IFNs produced by both pathways. To verify that the observed phenotype was a manifestation of aberrant MDA5 signaling, we tested Mavs−/− hMDA5 R822Q Tg and Ifnar1−/− hMDA5 R822Q Tg mice. Mice deficient for MAVS or IFNAR in the hM-R822Q Tg background showed decreased basal IFN levels in the small intestine (Fig. 4A). There was 100% survival of female Mavs−/− and Ifnar1−/− mice possessing the hM-R822Q transgene postinjection with poly(I:C) (Fig. 4B). In addition, no IL-6 was detected in the serum of Mavs−/− hM-R822Q and Ifnar1−/− hM-R822Q Tg mice with or without injection of poly(I:C) (Fig. 4C).
Similar to WT mice, histological sections of the ileum of Mavs−/− hM-R822Q mice had little evidence of poly(I:C)-induced damage (Fig. 4D). Damage in Ifnar1−/− hM-R822Q mice was also significantly reduced (Fig. 4D). The mRNA levels of Ifnb, Il1b, and Tnfa in the small intestines of both Mavs−/− and Ifnar1−/− hM-R822Q mice were comparable with those in WT (Fig. 4E). Meanwhile, the expression levels of poly(I:C)-induced Il6 mRNA in Mavs−/− and Ifnar1−/− hM-R822Q mice were reduced by 7-fold and 11-fold, respectively, compared with that induced in hM-R822Q mice.
Collectively, these data confirm that the inflammatory phenotype conferred by the hM-R822Q transgene was dependent on MAVS signaling. IFN signaling and ISGs also played a critical role, not only as effectors of inflammation but also for their subsequent function in increasing IFN expression. Furthermore, despite the ability of poly(I:C) to activate TLR3, the absence of inflammatory signature at the basal level because of deficiency in either MAVS or IFNAR prevented the generation of severe inflammation.
JAK signaling is critical for organ damage and lethality of poly(I:C) treatment
To assess the role of IFN signaling in the progression and potential prevention of disease, we administered the commercially available JAK inhibitor tofacitinib to female mice following the flow diagram shown in (Fig. 5A. After 3 d of daily oral gavage with tofacitinib or vehicle control, gavage on day 4 was followed by poly(I:C) injection into WT and hM-R822Q Tg mice. hM-R822Q mice treated with vehicle control died within 48 h of poly(I:C) injection. In contrast, hM-R822Q mice treated with tofacitinib and WT mice treated with vehicle or tofacitinib achieved 100% survival (Fig. 5B). There was also a 6-fold reduction in the serum levels of IL-6 in tofacitinib-treated hM-R822Q mice compared with vehicle-treated mice 20 h postinjection with poly(I:C) (Fig. 5C). After poly(I:C) injection, serum IL-6 levels in tofacitinib-treated hM-R822Q Tg mice were reduced to levels statistically comparable with those in WT mice (Fig. 5C). This corresponded to the absence of poly(I:C)-induced bowel distension in tofacitinib-treated poly(I:C)-injected hM-R822Q mice (Supplemental Fig. 3B). Villous lifting and epithelial shedding in the poly(I:C)-injected hM-R822Q mice were ameliorated by tofacitinib administration (Fig. 5D). Corresponding mRNA analysis of the small intestine showed reduced transcript levels of Ifnb, Il6, and Il1b in hM-R822Q mice when treated with the JAK inhibitor before poly(I:C) injection (Fig. 5E). Tofacitinib was also able to reduce inflammatory gene expression in nonmucosal organs such as the liver (Supplemental Fig. 3D).
These data show that blockade of JAK signaling effectively reduced the overreactive inflammatory response in hM-R822Q Tg mice to a degree that allowed survival. Damage at the organ level was reduced, and cytokine stormlike symptoms were abrogated.
Overexpression of MDA5 in mice has been shown to yield a chronic type I IFN signature. Crampton et al. (22) described mice with 12 copies of MDA5 having systemic elevated expression of ISGs, slightly enlarged spleens with immune cell populations comparable with WT mice, but with no overt signs of pathology. In our study, mice with constitutively active mutant human MDA5 R822Q displayed systemic upregulation of inflammatory cytokines and significant immune cell activation, with some histological pathology in the kidneys, spleen, and liver. This implies that the higher level of basal immune activity sensitizes hM-R822Q Tg mice to inflammatory stimulation. Consistent with this notion, the mucosal surfaces such as the intestines, eyes, and lungs showed the most apparent signs of vulnerability, even in specific pathogen-free conditions. Due to their roles in food absorption, sensory function, and gas exchange, barriers at these surfaces are thin and highly permeable. Even in the absence of active infection, natural microbiota are encountered at these sites, causing chronic low-level immune activation. In the presence of strong immune stimulants such as poly(I:C), the response shifts from low-grade chronic inflammation to massive systemic hypercytokinemia, resulting in significant damage to the intestines.
It must be noted that there was high expression of human and mouse MDA5 in hM-R822Q mice (Supplemental Fig. 1A, 1B). A large part of this may be credited to IFN-induced upregulation of MDA5 expression (32). Owing to the presence of the MDA5 promoter to the inserted human MDA5, both mouse and human MDA5 expression were induced by the IFN response produced by the constitutively active MDA5. Previous studies have shown that, in mice, highly elevated expression of WT MDA5 alone did not cause elevated cytokine response or produce any observable pathologies (22). In contrast with the increased lethality and excessive cytokine response in hM-R822Q Tg mice, overexpression of WT mouse MDA5 was found to elevate IFN-I response and increase resistance to viral infection (22).
MDA5 R822Q is found in patients with SMS, which typically manifests with aortic calcification, skeletal and dental abnormalities, and in some cases, glaucoma and recurrent infections. In line with this tendency, hM-R822Q Tg mice displayed abnormalities in bone structure and osteoclast development. Despite the absence of abnormalities in dentition (data not shown), the inflammatory signature in the mRNA profile, as well as the observed impairment of osteoclast differentiation, demonstrated conditions that are precursory to abnormal tooth eruption and tooth movement (33). Likewise, normal internal eye pressure indicated absence of glaucoma. However, inflammation in the eye region after poly(I:C) injection indicates the eye as a vulnerable organ in hM-R822Q Tg mice. The heart also showed strong SMS-associated pathogenesis. Although valve enlargement and fibrosis in the heart were only slightly increased, elevated calcification of the aorta indicated a high risk for dysfunction. This predisposition of hM-R822Q Tg mice to vascular pathology was supported by mRNA expression data. In particular, TNF-α, IL-1β, and IL-6 upregulation were found to have a causative correlation with aortic valve and vascular mineralization (27, 34–38).
One possible consequence of these heart and vascular abnormalities in hM-R822Q Tg mice may be shorter life spans compared with their WT C57BL/6J counterparts. Although symptoms of nephritis did not progress as the hM-R822Q mice aged, skeletal abnormalities, as indicated by vertebral posture, were more prominent in the Tg mice than in WT mice. We speculate that the chronic inflammatory signature in the heart and vessels also lead to progressive calcification of the aorta and aortic valves, which, together with the accumulation of other inflammation-induced organ damage, may contribute to early death in hM-R822Q Tg mice.
A point that piqued our interest was the extreme difference in response of male and female Tg mice to poly(I:C) injection. Poly(I:C)-induced lethality was observed in female hM-822Q Tg mice, but not in male mice (Supplemental Fig. 2A). In humans, there is generally an increased incidence of autoimmune disease in women compared with men (39–41). This is coupled with a stronger immune response and Ab production in women during infection and vaccination (42–44). Most studies have attributed this enhanced immune response to X-linked factors and hormones, such as estrogen (45–47). Meanwhile, no direct reports of sex differences in SMS have been made. After immune stimulation, IFN-β and IL-6 expression were more strongly induced in females, amplified by positive feedback from already activated innate and adaptive immune cells. Immune regulation in male mice disrupted this feedback mechanism and allowed inflammation to be resolved. Our data showed that male hM-R822Q Tg mice had increased IFN-β and IL-6 expression 20 h postinjection with poly(I:C) compared with uninjected mice, but this expression was still significantly lower than that observed in poly(I:C)-injected female hM-R822Q Tg mice (Supplemental Fig. 2E). Notably, however, BALB/c hM-R822Q mice did not exhibit lethality or any sex-specific differences in pathology (Supplemental Fig. 4D). This may be because of the significantly lower IFN and cytokine expression in poly(I:C)-injected BALB/c hM-R822Q Tg mice (Supplemental Fig. 4E). This further emphasizes the role of hypercytokinemia in lethality. Various studies have shown that genetic background plays a role in response to infection and susceptibility to autoimmune disease (48–52). A closer look at these disease-associated mutations in mice across multiple backgrounds may help elucidate some genetic basis for sex differences in the development of autoimmune disorders.
The absence of MAVS or IFNAR in female hM-R822Q mice also prevented the lethal cytokine storm induced by poly(I:C) injection. MAVS deficiency effectively eliminated MDA5-mediated signaling, also cutting off the signal immediately downstream of the Tg human MDA5 R822Q. Meanwhile, rescue of mice by ablation of the gene encoding the IFNAR underscores the importance of positive feedback and the signal amplification generated by IFNAR-mediated upregulation of MDA5 production. This is supported by the abrogation of systemic IL-6 overexpression in Ifnar1−/− hM-R822Q mice. The impact of signal amplification through IFNAR signaling is critical for disease development.
Likewise, inhibition of JAK/STAT signaling was able to disrupt type I IFN signaling. JAK inhibitors are being tested as treatment for various autoimmune diseases, such as systemic lupus erythematosus (53, 54), rheumatoid arthritis (55), and IBD (56). Their potential use in acute infection is also being widely explored. Baricitinib, for example, has been approved for emergency use against severe acute respiratory syndrome coronavirus 2 (57, 58). In this study, we used tofacitinib, which is a JAK1/2/3 inhibitor approved for use against rheumatoid arthritis, psoriatic arthritis, ulcerative colitis, and polyarticular course juvenile idiopathic arthritis (58). A similar study assessing tofacitinib efficacy against Con A–induced immune-mediated liver diseases showed that tofacitinib decreased serum levels of IL-6, TNF-α, and IFN-γ (59). Liver inflammation was visibly decreased based on liver appearance and histological analysis. This suggests that although drug treatment is unable to rescue mice from developmental damage, it can prevent inflammation-associated pathogenesis, including late-stage vascular calcification and liver damage. Our study has shown that it is effective in the short term for the suppression of otherwise potentially lethal inflammation. Despite this, we understand the limitations and risks of therapeutic approaches targeting other elements in the inflammatory pathways. Although effective in the short term, treatments using JAK inhibitors open the possibility of nonspecific and often adverse effects. The direct use of the human MDA5 in this mouse model may help facilitate further studies that specifically target the MDA5 protein, making drug treatments more tailored to MDA5-mediated disease and possibly translatable to human patients.
With our current data, we can conclude that the introduction of constitutively active MDA5 R822Q into mice leads to persistent, low-grade systemic inflammation that leads to progressive heart and bone abnormalities. This is comparable with a slow development of critical SMS-like symptoms. Furthermore, the mutant transgene caused a hypersensitive reaction to immune stimulation in mice, leading to lethal inflammation. This result suggests that the hM-R822Q genetic background presents a heightened risk for severe disease in the case of vaccination, particularly with RNA-based vaccines. This acute pathogenic response can be ameliorated through immune-suppressive treatment. Overall, hM-R822Q Tg mice show potential as an effective model for understanding both genetic and environmental factors that may contribute to the progression of disease in type I IFNopathies and may be instrumental in establishing clinical treatments for SMS-related spontaneous and induced pathogenesis.
We thank Dr. Shizuo Akira for kindly gifting the Mavs−/− mice. We also thank Dr. Tetsuo Noda for generating the Ifih1gs/+ mice. We also thank Prof. James Hejna for invaluable comments during the writing of this manuscript.
This work was supported by independent grants from the Japan Science and Technology Agency; the Ministry of Education, Culture, Sports, Science and Technology of Japan (innovative areas, infection competency, Grant 24115004); Japan Agency for Medical Research and Development under Grants JP17ek0109100h0003 and JP18ek0109387h0001; The Kato Memorial Trust for Nambyo Research; and the Japan Society for the Promotion of Science Core to Core Program. It was also supported by the Deutsche Forschungsgemeinschaft (German Research Foundation) under Germany’s Excellence Strategy (EXC2151; Grants 390873048 and TRR237; Grant 369799452, Project 404459591).
The online version of this article contains supplemental material.
Abbreviations used in this article:
bacterial artificial chromosome
human MDA5 bearing the SMS-associated mutation R822Q
inflammatory bowel disease
- Ifih1gs/+ mice
mice harboring the single missense mutation G821S in Ifih1
mitochondrial antiviral signaling
melanoma differentiation–associated protein 5
retinoic acid–inducible gene I
retinoic acid–inducible gene I–like receptor
telomerase repeat amplification protocol
The authors have no financial conflicts of interest.